Development Of DSP-based Instant Noodle Production Line Intelligent Detection Algorithm

Most instant noodle manufacturers currently detect potential missing noodle,flavoring or fork manually or using PC during manufacturing process.However,manual work will significantly increase labor cost and using PC method has disadvantages like large volume,high power consumption and weak resistance to interference.So developing a flexible,stable and low-cost detection system makes important sense for instant noodle manufacturers.This research develops an intelligent instant noodle detection system based on TMS320DM8127 DSP to accurately identify missing noodle,flavoring or fork.Based on computer vision,the research performs detailed and comprehensive analysis on features including texture,color,and shape and then develops appropriate algorithms for different objects which are successfully implemented on DSP core of DM8127 by extending EMCV library.In order to improve performance,the data transmission and storage modules are appropriately configured and the algorithm is optimized from several aspects such as compiler,C code and Cache.In addition,the system uses PC as backstage for controlling,sending parameter,displaying results and generating alert.After testing,the developed detection system could well satisfy functional and performance requirements during instant noodle producing process.The algorithm takes less than 500ms to detect one subject and the accuracy rate is over 99%.The system can adapt to complex detection background and light-variant environment of production line.